Methods for Constructing Normalised Reference Scores

The use of the lambda-mu-sigma (LMS) method for estimating centiles and producing reference ranges has received much interest in clinical practice, especially for assessing growth in childhood. However, this method may not be directly applicable where measures are based on a score calculated from question response categories that is bounded within finite intervals, for example, in psychometrics. In such cases, the main assumption of normality of the conditional distribution of the transformed response measurement is violated due to the presence of ceiling (and floor) effects, leading to biased fitted centiles when derived using the common LMS method. This paper describes the methodology for constructing reference intervals when the response variable is bounded and explores different distribution families for the centile estimation, using a score derived from a parent-completed assessment of cognitive and language development in 24 month-old children. Results indicated that the z-scores, and thus the extracted centiles, improved when kurtosis was also modeled and that the ceiling effect was addressed with the use of the inflated binomial distribution. Therefore, the selection of the appropriate distribution when constructing centile curves is crucial

Binomial distribution · Ceiling (cloud) · Ceiling effect · Cognition · Covariate · Econometrics · Estimator · Item response theory · Kurtosis · Normality · Psychometrics · Statistics · Child and Animal Learning Development · Early Childhood Education and Development · Mathematics · Medicine · Psychology · Psychometric Methodologies and Testing

• Longitudinal performance trajectories of young female sprint runners

  Open Access•Michael Romann, Marie Javet et al.•Frontiers in Sports and Active…•2024

  Incluir item

• Ggplot2

  Open Access•Hadley Wickham•Ggplot2•2009

  Incluir item

• Multi-ethnic reference values for spirometry for the 3–95-yr age range

  Philip H Quanjer, Sanja Stanojevic et al.•European Respiratory Journal•2012

  Incluir item

• Strategies for reducing inequalities and improving developmental outcomes for young children in low-income and middle-income countries

  Open Access•Patrice L Engle, Lia C H Fernald et al.•The Lancet•2011

  Incluir item

• Generalized Additive Models for Location, Scale and Shape

  Open Access•Robert A Rigby, D M Stasinopoulos et al.•Applied Statistics•2005

  Incluir item

• Smoothing reference centile curves

  Open Access•Tim J Cole, P J Green•Statistics in Medicine•1992

  Incluir item

• Flexible smoothing with B-splines and penalties

  Paul H C Eilers, Brian D Marx•Statistical Science•1996

  Incluir item

• Estimating the Dimension of a Model

  Gideon Schwarz•The Annals of Statistics•1978

  Incluir item

• An Analysis of Transformations

  Open Access•George E P Box, David R Cox•Journal of the Royal Statistical…•1964

  Incluir item

• Early childhood development coming of age

  Open Access•Maureen M Black, Susan P Walker et al.•The Lancet•2017

  Incluir item

• Investigating Ceiling Effects in Longitudinal Data Analysis

  Lijuan Wang, Zhiyong Zhang et al.•Multivariate Behavioral Research•2008

  Incluir item